It stops for the calls that are yours
You set the line. A dollar amount, a customer tier, a kind of action. Anything past it pauses and comes to you with the evidence and a preview, where you already work.
Maria at Harborline was charged twice for June. Refunding the duplicate.
Refund of $2,150 is over your $500 limit
- Customer
- Harborline, since 2024
- Reason
- Charged twice for June, Maria flagged it by email
- Rule
- Refunds over $500 go to Priya
Trust & control
Your line Its discipline
You decide what it may do alone, what it must ask about and what it may never touch. It holds to that on every run, and logs every step.
Permissions
You set the line. org.run stays on its side of it.
- Reply to routine support ticketsAuto
- Update CRM and customer recordsAuto
- Send invoice remindersAuto
- Issue a refund over $500Ask me
- Send a contract or proposalAsk me
- Pay a billAsk me
- Delete customer dataNever
Audit trail
- 09:14
Zendesk
Replied to ticket #2208
- 09:12
HubSpot
Updated Maple & Co. record
- 09:09
Stripe
Refund $640 — held for your approval
Scoped permissions
It only ever touches what you grant. Permissions mirror your org, down to the system, action and dollar threshold.
Complete audit trail
Every read, decision and write is logged with full context. Nothing happens off the record — ever.
Learns how you decide
Your approvals, edits and rejections teach it how you'd handle each call — so it asks less and gets more right over time.
Human approval gates
Set the line. Anything past it stops for your sign-off — payments, sends, deploys, anything high-stakes.
How it works
Decide in chat, dig in on the platform
- 01
You set the line
Refunds over $500, discounts over 10%, any production deploy. Each kind of action gets auto, ask or never.
- 02
It pauses with context
When a run hits your line, it stops and prepares a one-screen summary: what triggered it, the rule that paused it, and a preview.
- 03
You decide in seconds
Approve, edit or reject from Slack or Teams. Bayfield wants 15% off, your limit is 10%, and the counter is already drafted.
- 04
It learns from your call
Each approval, edit or rejection changes how it handles the next one, so it asks less over time.
Where you approve
Approvals come to you
Most decisions should take seconds and happen where you already work. The bigger ones ask for a little more.
Slack and Teams
A one-screen summary with approve, edit and reject buttons, in a DM or the channel you choose. Most approvals end here.
The org.run web app
Every decision, with the full run behind it. Pending approvals, the activity feed, your rules and the company map in one place.
The mobile app
Approve from your phone, with sign-in for the high-stakes calls, wherever you are.
Off your plate
Stop being the bottleneck on every yes
You still make the calls that matter. You stop chasing the context needed to make them.
- Digging through threads before you can say yes
- Approving routine items one email at a time
- Chasing who is waiting on your sign-off
- Rebuilding context on a deal you have not seen
- Re-checking the numbers someone sent you
- Hunting for what was approved and when
- Covering approvals while someone is on leave
- Answering the same exception the same way
Built for fast, safe decisions
Everything you need to make the call
One-tap approve, edit or reject
Decide from the message itself. Edit the draft in place when it is nearly right.
Deep link to the full run
Every message opens the full run: the trigger, the evidence, the rule that paused it, the preview and its history.
Risk-threshold step-up
Above a risk threshold, a chat button is not enough. Approval moves to the org.run app or asks you to sign in again.
Home inbox
Pending decisions, an activity feed of what it did alone, and what is waiting on whom, on one home screen.
Rules view
See every rule behind every pause, who owns it and how confident it is, and change it in plain language.
Routed to your own team
Each call goes to whoever owns it on your team: refunds to Priya, deploys to the on-call engineer. Your people act on the high-stakes steps, inside the run.
Need hands, not a decision?
For manual work nobody on your team should be doing, Human Hand sends one of our own experts.
See Human HandBatch approvals
Approve a run of routine items in one go, with the outliers pulled out for a closer look.
Approval history
Every yes, edit and no is kept with who decided and when, ready for an audit.
Questions about approvals
Your rules. Each kind of action is set to auto, ask or never, with thresholds like a dollar amount or a customer tier. One-way actions such as payments, sends and deploys are set to ask by default.
Yes. You get a one-screen summary with approve, edit and reject buttons. Above a risk threshold, approval moves to the org.run app or asks you to sign in again, because a chat button is too easy to press by accident.
What triggered the run, the evidence it used, the rule that paused it, and a preview of exactly what it will do: the email, the refund, the code diff.
Approvals follow your org chart. You can delegate a kind of approval to someone else while you are on leave, and every delegated decision is logged.
Less and less. Your approvals, edits and rejections change how confident it is in each rule, and you can raise the autonomy level for a kind of action whenever you are ready.
Early access
Keep the calls that are yours
Set your lines and let org.run handle everything below them.
Scoped, auditable and reversible from day one. Your data stays yours.
